Reaching out to mentors for "Gesture Control with OpenVINO"

[ebeetles]

Hi mentors!

I'm Elwin, a statistics and machine learning freshman at CMU. I am interested in the Gesture Control with OpenVINO project, especially because I have experience using the mediapipe gesture recognizer in projects, and have done computer vision research using the mediapipe facial landmark detector. However, I don't have much experience with the internal workings of the mediapipe models, and I was wondering if I should look into that for this project.

Also, is it possible for a more in depth description of what this project entails and what I should read or familiarize myself with to get started?

Thank you so much!

[adrianboguszewski]

@openvino-dev-samples @zhuo-yoyowz

[openvino-dev-samples]

hi @ebeetles

From my perspective, there are 3 key works you should pay attention to:

1. Study the detailed workflow of gesture recognizer in Mediapipe, and find out full model list used in this workflow.
2. Learn how to convert these models from TF lite to OpenVINO format.
3. Reconstruct the of gesture recognizer pipeline with OpenVINO, or build OpenVINO wrapper to replace these models in original pipeline.

Finally you can consider to design your gesture control application/demo.